random runs greater than or equal to the inferred value (oc=0.1). Accuracy was defined from the binomial 95% confidence interval: p +/- accuracy (Strauss, 1982). Growth predictions were determined from linear regression using the general linear model (PROC GLM) (Yang et al. 2006). The multiple regression model selection procedures R- squared, Backward Elimination, and Mallow Cp were used to determine the combination of indicators for prediction of each variable. The results from regression analysis were based on best model selection criteria of minimizing Mallow Cp and maximizing R2 and included only those indicators having a biological significance level of p < 0.05 (SAS, 2002). Results Ecological Classification The Pt. Washington restoration site can be classified ecologically as a wet pine flatwoods subtype of the coastal pine flat based upon the results from Canonical Correspondence Analysis (CCA) ordination, indicator species analysis (IndVal), and pre-harvest stand data. Canonical Correspondence Analysis ordination indicated the maj ority of the plots measured at Pt. Washington fall in between the environmental patterns (moisture05, pH, SOM) (Table 5-1) for mesic flatwoods and wet flatwoods measured at the reference sites (Figure 5-1). Indicator species analysis produced results showing that gallberry was the indicator for both wet flatwoods and the Pt. Washington restoration site (Table 5-2). When the data were analyzed by age class, the ordination produced the same vectors of moisture05, soil pH, and soil organic matter content (SOM), but with stronger results.(Table 5-3). CCA ordination did not show any clear separation along age class (Figure 5-2). Indicator species analysis did produce results showing that the restoration site had similar plant species as the young age class of the reference sites (Table 5-4). Witch grass and blue stem grass were species indicators for the young age class, while witch grass and wiregrass were found to be the species indicators of the Pt. Washington restoration site